In the 4 decades since Dr Averill A. Liebow introduced necrotizing sarcoid granulomatosis, there have been publications of numerous cases, but its nature and possible relationship to classical and nodular sarcoidosis have been and remain controversial. Liebow introduced necrotizing sarcoid granulomatosis as a provisional diagnostic term and stated that “the problem is whether the disease represents necrotizing angiitis with sarcoid reaction, or sarcoidosis with necrosis of the granulomas and of the vessels.” There has, as yet, been no definitive answer to the questions that he raised.
To determine whether there is a relationship between necrotizing sarcoid granulomatosis and nodular sarcoidosis in order to ascertain whether the current prevailing opinion that they are related is correct.
The world's literature on necrotizing sarcoid granulomatosis from 1973 to 2013 and nodular sarcoidosis from 1952 to 2013 was critically reviewed. One hundred three cases of necrotizing sarcoid granulomatosis and 111 cases of nodular sarcoidosis were found suitable for individual case analysis.
The data showed a striking overlap in the clinical, radiologic, and pathologic features of both entities, strongly supporting the conclusion that necrotizing sarcoid granulomatosis is a previously unrecognized manifestation of sarcoidosis and is essentially the same as nodular sarcoidosis. It is proposed that use of necrotizing sarcoid granulomatosis as a diagnostic term be discontinued and replaced by sarcoidosis with necrotizing sarcoid granulomatosis pattern provided that an infectious etiology can be reasonably excluded. Our concept of sarcoidosis should now be expanded to recognize that there is a continuous spectrum of necrosis ranging from minimal to extensive.
In 1973, Dr Averill A. Liebow reported a series of 11 young to middle-aged adults, the majority of whom presented with respiratory symptoms and nodular lung lesions.1 The pulmonary histopathologic findings consisted of granulomas that were often confluent, necrosis of variable extent within granulomas, and vasculitis (granulomatous, giant cell, and lymphocytic). Lymphocytic vasculitis occurring alone was unusual. Smears stained for acid-fast bacilli and fungi as well as cultures were negative. The clinical course was described as generally favorable. Dr Liebow provisionally termed this constellation of findings necrotizing sarcoid granulomatosis (NSG) and stated that “the problem is whether the disease represents necrotizing angiitis with sarcoid reaction, or sarcoidosis with necrosis of the granulomas and of the vessels.” 1 In the intervening 4 decades, his questions have not been definitively answered, despite the publication of several large series of patients stated to have NSG1–8 and numerous reports of individual and small numbers of cases.9–49 Four decades later, despite the large number of publications, the nature of NSG still remains controversial, although prevailing current opinion favors a relationship to nodular sarcoidosis (NS). The opinions expressed on the nature of NSG by the authors of series of cases are shown in Table 1.
Because NSG and NS present with nodular lung lesions and have markedly overlapping pathologic and clinical findings, it seems possible, even likely, that they are closely related, if not identical. In order to further explore this hypothesis, an in-depth critical review of the world's literature on NSG1–49 and NS50–101 was undertaken.
MATERIALS AND METHODS
The following definitions are used:
Necrotizing sarcoid granulomatosis: a lesion containing granulomas accompanied by variable amounts of necrosis and granulomatous vasculitis (Figure 1, A through D) with no evidence to support an infectious etiology. Granulomatous and giant cell vasculitis as described by Liebow1 are considered together as granulomatous vasculitis. Lymphocytic vasculitis as the sole vascular abnormality is exceedingly rare. Radiographic presentation is almost always described as nodules or masses.
Nodular sarcoidosis: radiographic presentation with nodules or masses seen on plain chest films and often confirmed by computed tomography (CT) scans (Figure 2, A and B). These lesions may simulate the appearance of metastatic or primary lung neoplasms. Unfortunately, there is no standard for the minimum size of a nodule or mass required for the diagnosis of NS. Only 10 of the publications on NS mention nodule size59,61–63,77,93,96–99 ; these range from 5 mm to 5 cm. The largest published series contains 33 cases.96 In 3 of the cases the nodules were less than 3 cm, and in the remaining 30 they were equal to or greater than 3 cm. There is no evidence to support an infectious etiology, and pathologic, clinical, and laboratory findings are consistent with sarcoidosis. The lesions consist of epithelioid granulomas that may or may not exhibit necrosis.
Classical sarcoidosis: sarcoidosis with necrosis that is either absent or minimal in amount. The term classical is introduced to differentiate what most currently regard as sarcoidosis from NS and NSG.
Using the search terms “necrotizing sarcoid granulomatosis” and “nodular sarcoidosis,” I retrieved a total of 58 publications1–58 reporting 179 cases of NSG and 43 publications59–101 reporting 102 cases of NS from a Medline search of the worldwide medical literature. The publications span the period from 1952 through 2013. The NSG publications that were reviewed included 8 abstracts, and those reviewed on NS contained 15 abstracts. Limited data were extracted from these abstracts written in English because of inability to translate the languages of the full publications.
Six NSG publications encompassing 8 cases were rejected for this study for the following reasons: did not meet the histologic criteria for NSG36,46 ; tuberculosis was the likely diagnosis30,47 ; no pathologic examination of lung tissue and no convincing evidence of cutaneous vasculitis40 ; the diagnosis presented was “polyarteritis overlap syndrome.” 26
The publication by Chittock et al6 presented 7 patients stated to have NSG, but vasculitis was said to be present in only 4. Except for 1 of the 4 examples of vasculitis, which was granulomatous vasculitis, the descriptions were not convincing and left the diagnosis of vasculitis in doubt. Furthermore, the case showing granulomatous vasculitis was not identified. This series appeared to contain 6 cases of NS and 1 of NSG, but which cases were NS and which were NSG could not be ascertained from the data presented, and therefore these 7 cases were not included in the data analysis.
1n 1979 Churg et al2 published 12 cases of NSG for which data were available for each case. An addendum at the end of the paper added another 9 cases and stated that those cases were similar to the previously reported 12 but provided no data for individual cases. In 1983 Churg's5 publication entitled “Pulmonary Angiitis and Granulomatosis Revisited” included a discussion of NSG that included 32 cases, 12 of which were the cases reported2 in 1979, 9 of which were the cases reported in the 1979 addendum, and 11 of which were cases that had not been previously reported. This was a brief discussion of the topic that did not give data for individual cases. Therefore, 20 of the cases published by Churg were not suitable for use in the analysis of individual cases.
Saldana3 reported 24 cases of NSG in abstract form only. Because data were not presented for individual cases, these 24 cases were not suitable for use in the analysis of individual cases.
There were no English abstracts available for 8 of the NSG publications that reported 8 cases in languages that could not be translated. Therefore, no usable data were available from those publications.*
Therefore, this publication reviews and analyzes 103 individual cases of NSG and 111 individual cases of NS. The findings from the publications that did not lend themselves to individual case analysis will be discussed below.
Because many of the publications did not specify the exact intrathoracic locations and laterality of enlarged lymph nodes, it was decided to aggregate enlarged lymph nodes anywhere within the thorax under the heading of intrathoracic lymphadenopathy.
The terms vasculitis and angiitis are used interchangeably and are equivalent.
I attempted to collect data for 52 data fields from each case. However, as is often true in this type of analysis, many of the publications did not address the items required for specific data fields. The data collected for each case were entered into an Excel spreadsheet (Microsoft Corp, Bellevue, Washington) from which the results were extracted.
Both NSG and NS exhibit a female predominance, with female to male ratios of 1.70 and 1.27, respectively. Necrotizing sarcoid granulomatosis occurs in a somewhat older age group than NS, with median ages of 42 and 35, respectively. Eighty-two percent of patients with NS are 40 years or younger, as compared with only 31% of NSG cases (Table 2).
The vast majority of patients with NSG (88%) and NS (84%) present with pulmonary and/or systemic symptoms. The systemic symptoms include fever, chest pain, chest tightness, weight loss, night sweats, fatigue, arthralgias, and malaise; the respiratory symptoms include dry cough, dyspnea, chest pain, chest tightness, and (rarely) hemoptysis (Table 3).
Extrapulmonary manifestations reported in both include eye, skin, liver, splenic, lacrimal gland, and central nervous system involvement. Central nervous system involvement is more frequently reported in NSG than in NS, but whether there is a statistically significant difference is undetermined. Sjögren or sicca syndrome and serum angiotensin-converting enzyme elevation are also reported to occur in both. It should be pointed out that because many of the publications do not mention either the presence or absence of the various extrapulmonary manifestations, no conclusions can be drawn as to whether there is a significant difference between NSG and NS in the incidence of extrapulmonary manifestations.
Intrathoracic Radiologic Features
Intrathoracic lymphadenopathy occurs in a significant number of patients in each category, but the frequency in NS (85%) is considerably higher than in NSG (33%). The majority of patients in both categories have bilateral involvement with nodules or masses. Single lesions are reported in 26% of cases of NSG compared with 20% in NS. Cavitation is reported in both categories, with a possibly greater incidence in NS. However, because many of the publications do not mention cavitation as being present or absent, it is not possible to be certain that there is a higher incidence of cavitation in NS. Pleural effusions are also reported in both categories (Table 4).
Techniques Used to Obtain Tissue for Pathologic Examination
Ninety-eight percent of cases of NSG were diagnosed using tissue obtained by surgical procedures including wedge resections, segmentectomies, and lobectomies. Two cases of NSG (2%) were initially diagnosed at autopsy.7 In NS, only 33% of cases were diagnosed using tissue obtained by surgical procedures. The remainder were diagnosed using transbronchial lung biopsy (TBLB) (35%), needle biopsy (9%), bronchial biopsy (2%), intrathoracic lymph node biopsy (8%), extrathoracic lymph node biopsy (3%), and a small number of miscellaneous biopsy procedures (Table 5).
Pathologists were authors or coauthors in 78% of the cases and 80% of the publications on NSG, as compared with 17% and 34% respectively for NS. By definition, granulomas were present in 100% of cases in both categories. Necrosis and vasculitis were present in 100% of NSG cases by definition. The necrosis varied from small to large in amount and was almost always present within the granulomas. Necrosis was described as coagulative or caseous, or was not described at all in many cases. In a few cases foci of infarction were said to be present and thought to be the result of granulomatous vasculitis. Suppurative necrosis within granulomas was rarely reported. Vasculitis was almost always granulomatous and involved both veins and arteries. Vasculitis was described as lymphocytic in just a few cases. Overall, the descriptions of the pathologic findings in NSG were found to be lacking in detail concerning the extent of necrosis, proportion of granulomas exhibiting necrosis, type of necrosis present, and whether infarction, if present, was associated with vascular occlusion due to granulomatous vasculitis. Necrosis was present in 14% of the cases of NS and vasculitis in 14%. None of the cases of NS exhibited both necrosis and vasculitis. In the cases for which results of staining for acid-fast bacilli and fungi and cultures were reported, there was no evidence of an infectious etiology. In a large number of cases there was no mention of cultures or microorganism stains (Table 6).
Follow-up and Outcome
Patients with NSG were followed for a median of 24 months, whereas those with NS were followed for a median of 6 months. Twenty-one patients with NSG received no treatment as compared with 24 for NS. Thirty-eight percent of NSG patients who received no treatment exhibited either complete or incomplete radiographic resolution of lesions on follow-up, compared with 84% for NS patients. Thirty-three patients with NSG and 59 with NS were treated with corticosteroids. Eighty-one percent of the corticosteroid-treated NSG patients and 73% of the corticosteroid-treated NS patients exhibited either complete or incomplete radiographic resolution of lesions. A very small number of NSG and NS patients were treated with corticosteroid and immunosuppressive drugs, but the small numbers preclude drawing any conclusions concerning outcome. Seven (7%) of the NSG patients died, whereas there were no reported deaths among the NS patients. The causes of death in the NSG patients are shown in Table 8. It is certain that NSG was the cause of death in 1 of the patients as the result of central nervous system involvement.8 It appears that the other 6 deaths were not the direct result of NSG. However, it is possible that in 1 of the cases in which NSG was initially diagnosed at autopsy and the cause of death was not stated7 death could have been directly related to NSG; the authors did not comment on this possibility.
In publications by Saldana,3 Churg et al,2 and Churg,5 there were a total of 44 cases of NSG from which data could not be extracted for individual patients. In the 24 cases reported by Saldana3 in abstract form only, the age range was 21 to 73 years, with a mean age of 49 years. There was female predominance, with a female to male ratio of 1.4:1. Two-thirds of the patients were asymptomatic and none had clinical evidence of extrapulmonary involvement. Lung nodules were unilateral in 88% of the cases. Sixteen of 18 patients were said to have been cured by surgery alone and 2 of 18 experienced recurrences following surgery. Resolution of the lesions was associated with corticosteroid therapy in 2 patients and chlorambucil in 1 patient (Tables 7 and 8).
The 1983 description of 32 cases of NSG by Churg5 included 12 cases that had previously been published2 ; data for each of those 12 cases are included in this study. The following apply to all 32 cases. There was marked female predominance, with a female to male ratio of 4:1. The age range was wide, with disease sometimes occurring in children or teenagers. About 25% of the patients were asymptomatic. Twenty-one of 29 patients had bilateral nodules, 1 had multiple unilateral nodules, and 7 had solitary nodules. Cavitation was occasionally seen. Hilar lymphadenopathy was present in 65% of the cases. Extrapulmonary involvement was present in 12% of the cases, with ocular involvement predominating. The series included 6 cases in which there was some hyalinization but no necrosis. The hyalinization might have been thought to represent a healed phase of necrosis, but there was no discussion of that possibility. Cultures and special stains for microorganisms were negative in all of the cases.
In order to consider the possible relationship of NSG and NS to each other and to classical sarcoidosis, a brief discussion of classical sarcoidosis follows. The following information is extracted from the 1999 American Thoracic Society's statement on sarcoidosis:
[Sarcoidosis] commonly affects young and middle-aged adults and frequently presents with bilateral hilar lymphadenopathy, pulmonary infiltration and ocular and skin lesions. The liver, spleen, lymph nodes, salivary glands, heart, nervous system, muscles, bones and other organs may be involved. . . . The disease shows a consistent predilection for adults less than 40 yr of age. . . . Most studies suggest a slightly higher disease rate for women. . . . The characteristic lesion of sarcoidosis is a discrete, compact, noncaseating epithelioid cell granuloma. . . . Granulomas may occasionally exhibit focal coagulative necrosis. . . . In the lung about 75% of the granulomas are located close to or within the connective tissue sheath of bronchioles and subpleural or perilobular spaces (a lymphatic distribution). Vascular involvement is observed in more than half of the patients with open lung biopsy or autopsy studies. . . . Nonspecific constitutional symptoms such as fever, fatigue, malaise and weight loss may occur in about one-third of patients. . . . Dyspnea, dry cough and chest pain occur in one-third to one-half of all patients. . . . Cutaneous involvement occurs in about 25% of all patients. . . . Ocular involvement may occur in 11 to 83 percent of patients. . . . Any part of the eye or orbit may be affected; uveitis is the most common of all sarcoid eye lesions. . . . Clinically recognizable involvement of the nervous system occurs in less than 10% of patients. . . . Spontaneous remissions occur in nearly two-thirds of patients, but the course is chronic and progressive in 10 to 30 percent. Serious extrapulmonary involvement (eg, cardiac, central nervous system, hepatic) occurs in 4 to 7 percent of patients . . . at presentation. . . . In most large published series, one-third to one-half of patients . . . were treated with corticosteroids. . . . Most patients stabilize or improve with treatment, but relapse occurs in 16 to 74 percent of patients as the amount of drug is tapered or discontinued. . . . At least 10 to 20% of patients sustain permanent sequelae (pulmonary or extrapulmonary). . . . Fatalities occur in 1 to 5 percent of patients, typically owing to progressive respiratory insufficiency or central nervous system or myocardial involvement.102
Two features of the pathologic findings in classical sarcoidosis deserve further elaboration: necrosis and vasculitis. Although the granulomas of classical sarcoidosis have traditionally been considered to be noncaseating or nonnecrotizing, necrosis is now well recognized as a feature of classical sarcoidosis. I prefer not to use the terms caseating or noncaseating in describing the microscopic appearance of granulomas, because these terms apply only to the macroscopically (grossly) visible cheeselike appearance of lesions seen in tuberculosis and other conditions. The terms necrotizing and nonnecrotizing are preferable when describing the microscopic appearance of granulomas. The first detailed description of necrosis in sarcoidosis was provided by Ricker and Clark103 in 1949. In reviewing material from 300 cases of sarcoidosis, they found varying amounts of necrosis in 35% of cases. These were described as minute areas of fibrinoid necrosis seen in the centers of individual granulomas. They noted that the greatest amount of necrosis was present in areas where granulomas coalesced. The finding of necrosis in the granulomas of classical sarcoidosis has been confirmed by several other investigators, who reported an incidence of necrosis ranging from 6% to 35% of cases.104–108 The necrosis has been variously described as fibrinoid, granular, eosinophilic granular, and coagulative. The necrosis in classical sarcoidosis is usually minute, spotty, and inconspicuous, involving the central portions of only a small percentage of granulomas. Rare cases of classical sarcoidosis may exhibit larger and even confluent areas of necrosis.36,108 Carlens et al109 noted an association of fever, erythema nodosum, and arthralgia with necrosis of mediastinal lymph nodes of patients with sarcoidosis. Zettergren110 found necrosis to be more prevalent in biopsy material obtained from patients with recent onset of sarcoidosis.
Granulomatous vasculitis/angiitis is a major histologic finding in classical sarcoidosis. In the period spanning 1944–1974, there appeared sporadic case reports of granulomatous vasculitis in classical sarcoidosis.111–120 In some of these reports the diagnosis of sarcoidosis is questionable.111,112,115,120 In 1976, Carrington et al107 reported the presence of granulomatous angiitis involving arteries, veins, or both in 42% of 47 open lung biopsy specimens from patients with classical sarcoidosis. In 1977, Rosen et al121 reported the presence of granulomatous angiitis in 69% of 128 open lung biopsy specimens from patients with classical sarcoidosis. Venous involvement was more prevalent than arterial involvement, and 61% of the specimens exhibited both venous and arterial involvement. It was pointed out that the presence and extent of granulomatous angiitis varies directly with the number of extravascular granulomas. Vasculitis is often difficult to recognize on hematoxylin-eosin–stained slides because the affected vessels are often located in the midst of confluent granulomas. The importance of elastic tissue stains in helping to identify these lesions was emphasized. A mean of 3.0 ± 2.0 slides were examined per case; 1.5 ± 0.35 stained for hematoxylin-eosin and 1.5 ± 0.35 stained for elastic tissue. A mean of 1.5 ± 0.35 different tissue blocks were examined per case. Six of the 128 open lung biopsy specimens (4.7%) exhibited confluent granulomas forming nodules, granulomatous angiitis, and necrosis, thus fulfilling the criteria for NSG.108 In 1984 Kitaichi and Izumi122 reported granulomatous involvement of small pulmonary arteries and veins in 53% of open lung biopsies from patients with classical sarcoidosis. A report by Yamaguchi et al123 in 1986 stated that granulomatous angiitis was detected in 36 of 75 (48%) cases of TBLB, with an average of 4.55 biopsy specimens per case. In 1991, Takemura et al124 reported finding granulomatous angiitis in 53% of TBLB specimens from 174 cases of classical sarcoidosis, with an average of 3.47 biopsy specimens per case. In 1992, Takemura et al125 reported on pulmonary vascular involvement in 40 autopsy cases of classical sarcoidosis. They found that granulomatous vascular involvement, either active or healed, was present in 100% of cases. A mean of 14 sections of lung per case was examined. Venous involvement was more prevalent than arterial involvement, and they also documented lymphatic vessel involvement in 90% of the cases. Pulmonary hypertension associated with granulomatous vasculitis appears to be rare, but a small number of case reports have appeared.126,127 There are also a small number of case reports of sarcoidosis simulating pulmonary veno-occlusive disease, apparently the result of granulomatous pulmonary vasculitis.128,129 Based on the foregoing, it seems reasonable to conclude that the real incidence of granulomatous pulmonary vasculitis/angiitis in classical sarcoidosis approaches 100%.
In 1952, 21 years prior to Liebow's1 publication on NSG, the first report of NS was published by 2 senior medical residents, McCord and Hyman.73 They described 2 cases of pulmonary sarcoidosis with the roentgenologic appearance of metastatic neoplasm. In 1958, Felson77 published 3 additional cases of patients with classical sarcoidosis who exhibited multinodular lesions on chest roentgenograms. In 1961, Schwarz130 reported that 12 of 77 patients with classical sarcoidosis exhibited atypical chest roentgenographic patterns that were not typical of sarcoidosis: multinodular and cystic. The number of multinodular cases was not stated. A 1970 report by Freundlich et al131 disclosed no cases with nodular infiltrates in a review of 300 chest roentgenograms from patients with proven classical sarcoidosis. A roentgenologic analysis of 150 patients with classical sarcoidosis published by Kirks et al132 in 1973 showed what they described as a “large nodular pattern” in 3 cases (2%). In 1973, Sharma et al89 reported 6 cases of NS in a series of 150 patients (4%). Römer,59 in 1977, reported 2 cases of NS in a series of 126 cases (1.6%). Therefore, the reported incidence of NS based solely on examination of plain chest radiographs ranges from 0% to 4%.
Nodules are a much more frequent finding in chest CT and high-resolution CT studies of patients with classical sarcoidosis. In 1996, Hashimoto et al133 reported that the most frequent chest CT findings in 35 patients were small nodules in 100% and irregularly thickened bronchovascular bundles in 90%. In a review134 of high-resolution CT published in 1998, it was stated that “the most frequent pattern consists of widespread small nodules, often with irregular margins, which typically have a bronchovascular distribution.” A study135 of 17 patients using high-resolution chest CT published in 2004 found small nodules measuring 2 to 5 mm in diameter in 100% of cases, typically distributed in peribronchovascular and subpleural lung areas. A 2013 report136 of high-resolution CT findings in 45 patients revealed micronodules in 91% and nodules in 82%. It should be emphasized that the current concept of NS is based primarily on the finding in plain chest films of relatively large nodules that can simulate the appearance of metastatic or primary neoplasms. If, in the future, the diagnostic criteria for NS should be expanded to include small nodular disease seen on CT studies, its incidence would increase up to or very close to 100%.
Based upon the data presented from analysis of individual cases of NSG and NS, it is apparent that there is striking degree of overlap of their clinical, radiologic, and pathologic manifestations (Table 9). Both exhibit a female predominance, and the vast majority of both are symptomatic upon presentation. Necrotizing sarcoid granulomatosis presents in a somewhat older age group than NS. They both exhibit extrapulmonary involvement, with eye involvement appearing to be the most common in each. Involvement of the skin, spleen, liver, lacrimal glands, and central nervous system is reported in both. There are more reported cases of central nervous system involvement in NSG than in NS. Intrathoracic lymphadenopathy occurs in both but is reported much more frequently in NS. Nodules or masses are the characteristic radiographic finding in both entities. Cavitation is reported in both but has been reported more frequently in NS. Instances of the Sjögren/sicca syndrome have been reported in both. Elevation of serum angiotensin-converting enzyme occurs in both but has been reported more frequently in cases of NS.
Necrosis and vasculitis are present in 100% of cases of NSG by definition but are also reported in 14% and 14% of cases of NS respectively. Lymphatic/lymphangitic distribution of granulomas has been reported in both. In those cases where stains and cultures for microorganisms were reported, there was no evidence to support an infectious etiology in either condition. Complete and incomplete radiographic resolution in untreated patients is seen in both, with an incidence of 38% in NSG and 84% in NS. Necrotizing sarcoid granulomatosis and NS are both reported to be corticosteroid responsive and to have an overall benign clinical course. There are no deaths reported in NS. Of the 7 deaths reported in NSG (Table 8) only 1 appears to be definitely due to NSG with involvement of the central nervous system.8 It is possible that NSG was the cause of death in the 1 case in which the cause of death was not stated.7 The other deaths do not appear to have been directly related to NSG.
Two significant differences between NSG and NS are how the diagnosis is made and whether or not there is pathologist involvement in the publications. One hundred percent of NSG cases are diagnosed by examination of tissue obtained by a surgical procedure, whereas only 33% of NS cases are diagnosed in this way. The reason(s) that many fewer surgical lung biopsies were performed in the NS cases was not apparent from this case analysis. In NS, 67% of the cases are diagnosed based upon examination of relatively small biopsy specimens, with TBLB being the most prevalent of the biopsy procedures in that group. Overall, the amount of tissue undergoing pathologic examination is considerably less in NS than it is in NSG. Therefore, it would be expected that the likelihood of finding necrosis and vasculitis should be much less in lung tissue examined from NS cases than it would be in NSG cases. As has been previously stated, the finding of granulomatous vasculitis in TBLB specimens is rare in the author's experience. The literature on NS reviewed for this publication reports only 1 case of granulomatous vasculitis83 among 41 cases with TBLB (2%). This is difficult to reconcile with the experience of Kitaichi and Izumi122 and Yamaguchi et al,123 who reported finding granulomatous angiitis in TBLB in 48% and 53% of cases respectively. There was no indication in those reports as to whether extensive serial sectioning of the biopsy tissue might have been done. The fact that pathologist participation in publication of NS cases (32%) was considerably less than it was in NSG (78%) makes it likely that both necrosis and vasculitis have been underestimated in NS. I believe that true incidence of necrosis and vasculitis in NS is higher than has been reported. Therefore, it seems reasonable to assume that if surgical lung biopsies had been performed in all cases of NS and had been carefully examined by pathologists for the presence of vasculitis, its incidence would approach the 69% incidence reported in open lung biopsy specimens by Rosen et al.121 Of course, the likelihood of finding necrosis and vasculitis is directly proportional to the amount of tissue examined. The autopsy study of classical sarcoidosis by Takemura et al124 in which granulomatous vasculitis was reported in 100% of cases strongly suggests that if the lesions of NS were extensively sampled, the incidence of granulomatous vasculitis would approach 100%.
Now, 4 decades after Dr Liebow1 suggested the provisional designation of NSG for the 11 patients that he reported, I believe that the accumulated data from the subsequent medical literature provide striking evidence to confirm the currently prevailing opinion that NSG is a previously unrecognized manifestation of sarcoidosis and is essentially the same as NS. The only differences between the two that have emerged from this study are that NSG occurs in a somewhat older age group, it exhibits a lower incidence of intrathoracic lymphadenopathy, and complete radiographic resolution in untreated cases appears less likely to occur. I believe that granulomatous vasculitis has been underreported in cases of NS that underwent surgical lung biopsy; only 4 of 37 cases (11%) were reported to have granulomatous vasculitis. This is much lower than the reported 69% and 54% incidence of granulomatous vasculitis reported in open lung biopsies.121,122 As previously stated, granulomatous vasculitis in classical sarcoidosis is often difficult to identify in hematoxylin-eosin–stained sections, especially if one is not searching for it. Elastic tissue stains greatly facilitate its recognition. The only indication of the possible prevalence of the NSG histopathologic pattern among individuals considered to have classical sarcoidosis was the finding by Rosen et al121 that 4.7% of 128 open lung biopsies from patients with classical sarcoidosis exhibited epithelioid granulomas with necrosis as well as granulomatous vasculitis.108
Because the evidence now indicates that it is almost certain that NSG is a manifestation of sarcoidosis, I propose that the use of NSG as a diagnostic term be discontinued and replaced by sarcoidosis with NSG pattern. Although it has been known for decades that the granulomas of classical sarcoidosis may exhibit necrosis in small amounts, our concept of the disease should now be expanded to recognize that there is a continuous spectrum of necrosis ranging from minimal to extensive. The separation of NS and NSG is probably completely arbitrary and can vary with sampling and time course of disease.
It is widely accepted that the pathologist cannot make a diagnosis of sarcoidosis based solely on tissue examination. Good communication among the pathologist, clinician, and radiologist is essential in order to arrive at the diagnosis of sarcoidosis, which is usually made by the clinician after correlating clinical, pathologic, radiologic, and laboratory data. The pathology report should state the findings, including necrosis and granulomatous vasculitis if present. The most important consideration in lesions exhibiting necrosis is to exclude an infectious etiology. Stains for acid-fast bacteria and fungi must always be done and the results reported. However, the sensitivity of tissue acid-fast stains is relatively low, and a negative acid-fast stain does not exclude the possibility of mycobacterial infection. Cultures for microorganisms are a very important supplement to acid-fast and fungal stains; thoracic surgeons must be encouraged to submit tissue for culture in all cases that are not obviously neoplastic. In cases where there is a large amount of necrosis and granulomatous vasculitis, the possibility of Wegener granulomatosis might come under consideration. However, the presence of many compact, well-formed epithelioid granulomas and granulomatous vasculitis are not typical features of Wegener granulomatosis, and therefore the distinction between the two usually does not present much difficulty.
Sarcoidosis is currently regarded as a disease in which individuals with a genetic predisposition to develop granulomas may do so in response to a variety of environmental antigenic exposures that act as triggering agents. The identity of the triggering agents and the factors responsible for the varied clinical, radiologic, and pathologic manifestations, including the nodular and NSG patterns, are currently unknown and have yet to be elucidated.
In 1973 Liebow1 expressed his uncertainty about the nature of NSG, stating that “the problem is whether the disease represents necrotizing angiitis with sarcoid reaction, or sarcoidosis with necrosis of the granulomas and of the vessels.” I believe that the data presented in this study provide convincing evidence that “sarcoidosis with necrosis of the granulomas and of the vessels” is the correct alternative.
The author wishes to express thanks and appreciation to David Botros, MD, Barbara Elish, MSLS, and Dina Mauro, AAS, for their assistance in obtaining the publications for this review; to Bridget Bransteitter, DO, Elie X. Gertner, MD, and Triona Henderson, MD, for contributing case material; and to Mitchell Rosen, MS, for assistance with data management.
The author has no relevant financial interest in the products or companies described in this article.